Improving monitoring of cardiac function in critical illness: from advanced echocardiographic technology to metabolic and biochemical biomarkers predicting myocardial dysfunction.

Chapter 1: Validation of a deep neural network which integrates segmentation and as well as deformation imaging analysis in a large image database of echocardiograms. The objective of this chapter is to optimize deep learning techniques through study of a modified neural network, currently in development by the team of Jan D’hooge. This modified neural network will integrate local morphology of the myocardium derived from automated segmentation, with strain analysis in echocardiography. Segmentation is the process of delineation of cardiac structures and is mainly manually conducted in clinical practice. Strain and strain rate are both deformation imaging modalities nowadays available on most commercially available ultrasound systems. Strain is the percentage of change in length during systole, and strain rate is the rate of this deformation. Strain and strain rate are still load-dependent but to a lesser extent compared to ejection fraction. I will validate this newly developed neural network against a large dataset provided by participation in a European multicenter study: CARdiomyopathy in type 2 Diabetes mellitus ‘Cardiateam’. Cardiateam aims to assess the uniqueness of diabetic cardiomyopathy (DCM) and its pathophysiology. Cardiateam will recruit 1600 patients which will be fully documented for clinical, biochemical including genetic analysis, echocardiography and cardiac MRI. All echocardiographic data of 1600 included patients will be centralized at the Lab on cardiovascular imaging and dynamics (part of the department of cardiovascular sciences) of KU Leuven, which is appointed as echocardiography core lab. I will analyze and delineate all received echocardiographic imaging data. Chapter 2: Shear wave elasticity imaging as a non-invasive hemodynamic monitoring and diagnostic tool for critically ill patients at the intensive care unit. The Objective of this Chapter 2 is to investigate if Shear Wave Elasticity Imaging is able to assess cardiac filling pressures in critically ill patients at the ICU.  Shear Wave Elasticity Imaging (SWEI) is a recently developed high frame rate echocardiographic imaging modality which measures shear waves (SW) emitted by the myocardium following a mechanical event (such as valve closure). Shear wave propagation speed is intrinsically determined by myocardial elasticity, and consequently reflects ventricular filling pressures given the end diastolic pressure volume relationship. Acquisition of shear waves on the interventricular septum is made possible with an experimental scanner (HD-PULSE) at 1050 +/- 220 frames per second through a parasternal long axis window. I hypothesize that shear wave propagation speed is correlated with invasively measured filling pressures in the critically ill patients. Secondary, I hypothesize that shear wave propagation speed is able to predict fluid responsiveness in critically ill patients when combined with ventilator strategies to alter preload based on heart-lung interactions. I will assess the ability of SWEI to detect subtle variations in loading conditions during the respiratory cycle, the correlation between SW propagion speed and pulmonary artery occlusion pressures and the ability of SW propagion speed to predict fluid responsiveness. Chapter 3: Impact of sepsis on the heart: study of histological and metabolic changes in a resuscitated septic mouse model. The objective of this chapter is to study histopathological changes in the heart, changes in preferred metabolic substrate by the myocardium and possible biomarkers for these alterations during the course of sepsis in a mouse model. A clear insight in the exact pathophysiology, time course and treatment of critical illness induced myocardial dysfunction is lacking. Clinical practice and also future scientific work would benefit from a better understanding of the mechanisms and the extent by which critical illness affects the myocardium. The sepsis induced mouse mode by cecal ligation and puncture (CLP) lends itself well to study the impact of critical illness on the myocardium. However, current data on the impact of critical illness on the myocardium in murine experiments is limited to the first 24 hours after induction of sepsis15. The extent by which sepsis has an impact on the heart in the subsequent subacute or chronic phases has not been studied so far. In a previously conducted study, a CLP mouse model was studied to investigate pathways of cholestasis in critical illness. They were resuscitated using fluids and treated with broad-spectrum antibiotics, followed by sacrifice on 30 hours, 3 days, 5 days and 7 days. Hearts were frozen at time of sacrifice and will be used in our present study to investigate metabolic and pathological changes during sepsis. This timeline study is expected to contribute to our understanding of the impact of sepsis on the heart in a mouse model during and after resuscitation and may orient future functional echocardiographic studies.